Don't call it the next Microsoft. If things go Bob Young's way, people will think of Red Hat as the Wal-Mart of the open source world.
The Cat in the Red Hat
His company’s wildly successful IPO kicked off the market’s Open Source infatuation. So what’s on Bob Young’s mind these days?
by Robert McMillan and Lee Gomes
|all photos © ladi novotny |
Depending on whom you ask these days, Red Hat is either the poster child of the open source revolution or an unstoppable juggernaut. Everyone agrees on one thing though: Whatever Red Hat does has a major impact on Linux. Since it was founded by Bob Young and CTO Mark Ewing almost six years ago, the North Carolina company has done more than simply prove that a successful business can be built on Linux — its amazing IPO sparked the financial market’s current love affair with open source.
But will Red Hat be able to live up to the market hype and, most importantly, can it balance the demands of a public company with those of the Linux community? Or does it even need to?
Linux Magazine’s editorial staff recently got together with Red Hat co-founder, chairman, and self-proclaimed “spokesmodel,” Bob Young, to find out where he sees his company going. In attendance were Publisher Adam Goodman, Executive Editor Robert McMillan, Editor-at-Large Wendy Goldman Rohm, and contributor Lee Gomes.
LINUX MAGAZINE: Why is Red Hat’s stock worth so much?
BOB YOUNG: There’s only one fundamental reason for this. Our investors genuinely believe that we have, in fact, a better model for developing infrastructure software than the proprietary binary-only model. Operating systems are endemic. Every computer requires one, every application needs one to run. So our investors are just projecting five years into the future and they’re saying: “These guys have this massive opportunity. If they’re able to execute on that opportunity, this thing’s going to be worth a lot of money.”
LM: You once said that you didn’t want Red Hat to make Microsoft-style monopoly profits. You wanted to make the operating-system business a $500 million commodity business. What did you mean by that?
BY: What we’re talking about is something like Federal Express. At the time they got into the business, delivering packages between cities overnight was a $200 service. Federal Express was not trying to take over that $200 service. They were trying to do a service worth 10 bucks a package. In doing that, they reinvented the need and they turned a sleepy, few-hundred-million-dollar-a-year industry into a multibillion-dollar industry because people found new applications for that service.
And that’s effectively what we’re betting. We’re betting on the other 5.6 billion people who don’t have personal computers.
LM: If that’s the case, what will Red Hat will be selling in five years that you’re not selling now? Will you need to branch into new lines of business, selling applications, for example?
BY: This whole industry is evolving very quickly, and the open source space of it is evolving faster than any other. The killer apps that are going to be driving our growth five years from now haven’t even been thought of today.
Having said that, there are some great indicators today of the kinds of things that are going to be happening in the future. For example, one of the great indications is on the Tivo set-top box. Basically it’s video on demand. What you get is a Red-Hat-Linux-based box with a cable connector in it that allows you to record any television show at any time. They’re recording every television show ever broadcast on their central server, which can upload it to your Tivo set-top box.
The reason Tivo picked Linux is that they can run the same operating system in this Internet appliance on top of your television as they run on their servers. Having the same set of APIs and the same development environment in both places reduces their training and development and engineering costs dramatically, and on top of that they get complete source code and a license to do whatever they need to do for the benefit of their application.
So when you look at the Tivo example and you start extrapolating what are the killer apps that are coming, it’s a pretty safe bet that other highly innovative teams are using open source technology because they get benefits that they simply don’t have with proprietary binary-only operating systems.
LM: Do you think that Red Hat will ever develop and sell proprietary software of any form?
BY: Forever is a long time.
LM: Within the next five years then?
BY: Put it this way, the moment we do, we limit our opportunities because we will have given up on the bigger opportunities available to us at the moment. The big opportunities are available to us as an open source company.
LM: If you could snap your fingers and port any application to Linux, what would that be?
BY: I can think of a dozen that are important. I can’t think of any one that would make an impact by itself. I can point at one application that I think would improve our market share by one percent, but in the end who cares if we go from 16 percent to 17 percent? We need to go from 16 to 40 percent. So one application that takes our market share up by one percent isn’t going to do it. I need 100 applications that are going to take our market share up by 20 percent. What’s important and what is exciting to me is the volume of applications being ported, not the single applications.
LM: For a second there I thought you were going to say Microsoft Office.
BY: Sure. It was certainly one of our early ambitions, although we used to say this facetiously. Our definition of success back in 1995 was to build enough market share that Microsoft would be forced to port Office to Linux. We no longer say that in jest. In my personal opinion, I still see this as one of our definitions of success. I now see it as simply a matter of time.
LM: Well, how much time?
BY: Oh, I learned long ago not to predict in time.
LM: Why did Red Hat acquire Cygnus Solutions?
BY: The attraction to Cygnus was marketing and technology. The interesting thing with Cygnus was negotiating with their management team and board of directors. They didn’t recognize the value of their own brand.
LM: What made you realize that?
BY: Revenue stream. Their revenue stream going forward on their open source tools. The core Apache developers own the Apache server as tightly under an open source license as they would have had they put it under proprietary license.
We own RPM and Red Hat as effectively, for the purpose of earning any real dollars off it, as tightly as if we had put a proprietary license around it. And Cygnus owns the compilers as effectively as if they’d never distributed source code.
LM: So what did you see at Cygnus that told you they hadn’t leveraged that ownership?
BY: Recently they’d begun to do that. Within the last year or so, they’ve begun packaging their commercial compilers and putting them through retail. But do you know the inspiration for that?
LM: They told us it was Red Hat.
BY: They were saying, “We put this software in retail and now we’re making a million dollars a year!”
LM: So what is the most important thing that Red Hat gets from Cygnus?
BY: Cygnus was the other major open source company. They build key pieces of technology that all of our customers rely on: the compilers and the libraries. So our ability to serve our major customers — people like Northern Telecom, Intel, Fujitsu, and Sony — was absolutely dependent on being able to deliver a complete operating-system environment with an identical set of APIs from the smallest embedded application to the largest servers. And combining Cygnus’ compiler and library expertise with Red Hat’s kernel and operating-system expertise just made so much sense.
LM: But one of the side effects of the Cygnus acquisition is that you now have two different sets of APIs, isn’t it? You now have the ECOS APIs as well as Linux?
BY: I don’t have good answers for you on ECOS other than that we’re very excited. Linux is a big monolithic kernel. It’s designed for native platforms, like PCs or servers. And it’s not small enough or realtime enough for many of the embedded applications out there. Of course there are a lot of embedded applications that aren’t as resource-constrained and don’t require realtime functionality, for which Linux makes a great embedded solution. But for many, Linux has limitations that ECOS addresses much more effectively.
I don’t think there’s anyone at Red Hat or Cygnus who can speak with a huge degree of certainty as to what the future of Linux versus ECOS is in the embedded space. And as near as we can tell in our preliminary view of things, we have to do both ECOS and Linux.
LM: Broadly speaking, is there a strategy to unify these APIs?
BY: I would duck the whole subject and encourage you to interview Michael Tiemann. Michael will have a very large ongoing role with the merged Red Hat/Cygnus company and, in fact, is driving the development of those strategies.
LM: So with all of this technology, is Red Hat going to become the next Microsoft?
BY: People keep accusing us of trying to be the Microsoft of the open source space, and I just genuinely get frustrated with that question because it implies a fundamental misunderstanding of how our business model works. We’re not in the proprietary-software business, so we can’t be the Microsoft of this space. We’re trying to be the Wal-Mart.
A lot of retailers don’t like Wal-Mart, and I’m sure that there will be lots of Linux vendors who won’t like Red Hat by virtue of our success. We eliminate opportunities that they otherwise might have had if we weren’t successful.
LM: We’ve heard that big companies like IBM are making investments in a wide variety of Linux companies because they don’t want any one of them — and usually you guys get mentioned here — to become the next Microsoft.
BY: Yeah. IBM are so schizophrenic. Allowing Microsoft to steal the OS opportunity on them and effectively handicapping them is the great turning point in IBM’s history. And they never want to make a mistake like that again.
So here is a new OS springing onto the scene. They could care less if it was open source and if it was made on Mars and whatever. They don’t want to see any one vendor control that operating system.
LM: Ignoring the Microsoft accusation, how do you feel about the perception that Red Hat is now somewhat above it all when it comes to the Linux community?
BY: I sure hope so. We’ve been above it all from the very outset. Our goal was never to compete with Slackware, even though Slackware had 95 percent of all the Linux users when we started. They had 95 percent of the half-million Linux users there were at the end of 1994. Today, the best estimate is there are one to two million Slackware users. So we haven’t actually converted any Slackware users to Red Hat. I’ve done a terrible job of doing what people accuse me of doing.
LM: That was actually the point: To what extent do you guys see the Linux community as your base? Some feel that Red Hat is somewhat disconnected from the ethos of the Linux community.
BY: I see where you’re going with this. You’re listening to the Anonymous Coward on Slashdot. When an Anonymous Coward on Slashdot posts some diatribe at the expense of Red Hat, we don’t worry about it.
When Linus Torvalds posts a concern about Red Hat, we pay a great deal of attention and we actively engage him in this and say: “Look. Either you’re wrong and you’re missing some data or we’re wrong and we’ll fix it.” But we don’t allow that kind of key influencer in this community to get very far apart.
LM: If you talk to these key influencers, none will say that they follow a company roadmap. They will all say the same thing: that they write code for the love of hacking and that they add code on the basis of technical merit. Is this lack of control a problem for a company like Red Hat?
BY: Alan Cox works for Red Hat. He takes feedback from our engineering team as to what SAP and Oracle are looking for in the next version of the kernel. He takes every bit as much feedback from what Stephen Tweedie, Dave Miller, and Linus think would be really cool to do next as well.
LM: What if Caldera wants something in the next kernel?
BY: That’s a little bit of the problem with the secondary Linux distributions; they simply can’t provide the technical roadmap because they aren’t building as much of the technology. They’re much more dependent on other people to build the technology for them. So Mandrake did this wonderful job from a marketing point of view but they’re getting zero traction in the enterprise because everyone knows that Mandrake is basically Red Hat with Mandrake’s name on it.
LM: To the same extent that Alan Cox is working for Red Hat, isn’t he also, by hacking for the love of hacking, working for Caldera?
BY: Yes. Absolutely. And it’s up to Caldera to take advantage of him. But the problem is that there’s this engineering component and then there’s a marketing component. And if you don’t own your product, then the competition in the marketplace isn’t on technology.
LM: What kind of impact do you think Windows 2000 will have on Linux’s momentum?
BY: It keeps coming back to this: Do you believe in what you’re doing or don’t you? If you don’t believe in what you’re doing then you think, “Oh boy. We’d better get to work and take advantage of the window of opportunity because when Windows 2000 comes up we’re in trouble.” I’m indifferent to Windows 2000 because we’re on a marathon. If it wasn’t a marathon, it wouldn’t be interesting.
Windows 2000 is one section of the marathon, one that you have to do well on, but you have to do well on everything else in order to win the damned marathon.
LM: What do you guess the technical reception to Windows 2000 will be?
BY: It’s going to be interesting technology. It’s not going to be that stable. It’s going to require a huge adoption curve; it’s going to give us a great opportunity because people are going to be faced with porting their whole environments from wherever they are over to this 2000 thing. But when you create a migration requirement people then start looking and saying: “Well if I have to do this and it’s going to be painful and costly, what are my choices?”
So it’s going to create opportunities for Novell and for us and for Sun. Microsoft has such market dominance that a series of bad reviews would slow their adoption by a few percentage points versus a series of good reviews. Unless their stuff is really crap and it has a material impact in the same way Windows NT was impacted by its early versions.
It’s like we’re at mile six or seven of a marathon and some guy has just started on a sprint. Well good for him. I’ve got a market position that I’ve got a great deal of confidence in.
LM: Do your engineers have access to beta Windows 2000 code that they can scrutinize?
BY: They’re a little too busy to do that. The key thing is that our engineers stay focused on open source. We are more focused on building the tools to support open source technology.
LM: What impact, if any, do you think the antitrust trial has had on your company?
BY: The bulk of the benefit has already been felt in the marketplace in that the deals that we’ve been able to put together — the relationships with Dell and Compaq and IBM, Oracle, SAP, and Intel — would not have been doable prior to the Justice Department investigating Microsoft.
What the Justice Department represents is a policeman on the beat. For example, there are laws against spitting in the street, but when was the last time someone was actually arrested for this? If you don’t enforce the laws, people will tend to ignore them. Well Microsoft are past masters at ignoring laws that are not enforced.
So the Justice Department investigation effectively puts a policeman on the street. Microsoft can’t call up Michael Dell and threaten him anymore because they know that that call will end up in the Justice Department’s hands within five minutes.
LM: What’s your relationship to the Dells and Compaqs of the world?
BY: Actually I talk to Michael Dell on a regular basis these days. Dell is almost an open source company.
LM: In what sense?
BY: In the hardware sense. There’s a big difference between Dell and Compaq, for example. Compaq does not really run Linux very well. Compaq keeps building proprietary enhancements to their hardware that the guys in the Linux community can’t take advantage of because they are proprietary to Compaq. Dell, on the other hand, says, “We’re just going to use off-the-shelf Intel PC technology and we’re just going to do it better than everyone else.”
LM: Why did you step down as CEO of Red Hat?
BY: Well, I don’t think of it as stepping down, and if you saw my workload you wouldn’t think of it as stepping down either. I was chairman and CEO, and Matthew Szulik was president and had all the operating responsibilities. So we shifted the CEO title sideways onto Matthew’s shoulders because Matthew is very, very effective at leading large organizations to build very efficient processes for serving large volumes of customers very consistently. Whereas my track record and my training is all around innovation: Identifying new market opportunities and innovating business models. So building processes is something that is not part of my chemical make-up.
LM: What do you think the Microsoft Windows story on Merced is?
BY: I couldn’t tell you. It’s one of these things where I wouldn’t want to be the Microsoft product manager.
There are certain jobs that you simply would not want. There are two jobs I would not want: I wouldn’t want to run SCO, and the other is I wouldn’t want to be the Microsoft product manager on Windows. I wouldn’t mind running Microsoft. That’d be a fun job.
LM: What would be the first thing you’d do?
BY: Running Microsoft?
BY: I’m not sure I could tell you guys even off the record.
Robert McMillan is executive editor of Linux Magazine. He can be reached at firstname.lastname@example.org. Lee Gomes writes for the Wall Street Journal.